the cognitive engine : artificial intelligence for wireless

Report
Instructor: Dr.George Collins
NIREESHA NAMBURU
 Cognitive radio architecture
 Cognitive engine design
 Components descriptions
1.sensors
2.optimizer
3.decison maker
4.policy engine
5.radio framework
6.user interface
7.cognitive controller configuration
 AI and its techniques
 Conclusion
 References
User domain
Operational cognitive radio platform
Cognitive
engine
Policy
engine
Policy
domain
Networking/
radio
communicat
ions system
Environmen
t & RF
channel
PSCR
CR GUI
User interface
Energy
detector
Signal class
WSGA
sensors
Cognitive
optimization
n
controller
Rule based
meters
Policy
verification
Decision
maker
Knowledge
base
SQL
database
Radio
framework
GNU radio
Fuzzy logic
WSGA verifier
XG
IRIS
KUAR
 Sensors collect data from radio or other systems to
describe and model the environment.
 The important aspect of a sensor is having a standard
approach to how data is transferred to a cognitive
controller.
 Application programming interface(API)
 Initialization
 Waiting for data request from cognitive controller
 Collecting data and building a model
 Transferring model to cognitive controller
 Cognitive engine sends
information to the sensor
through some generic
interface
 Sockets and SOAP :
communicate information
between software
programs
 Functions and processing
algorithms are retrieved
through external library
Sensor state machine
Sockets,
SOAP,etc
External
application
,library etc
Interface
description
Data collection
library
XML modelling
<?xml version=“1.0”?>
<sensor>
<model-name>”model-name”<\model-name>
<data-tag type=“type” size=“size” unit=“unit”>”value”<\datatag>
…….
<\sensor>
 the optimization process takes the user oriented information
from sensors or user interface to select or design a waveform
that will maximize the performance.
 Optimizer produces waveform that comes close to the QoS
values with respect to the provided environmental data.
 Depending on the implementation the optimizer may build a
new waveform or select it from a list of predefined waveforms.
 It coordinates information and decides how to optimize and act.
 If optimization is required the decision maker will provide
some context such as optimization goal or time limit for when a
new waveform is required.
 The current method of decision making is based on CBDT.
 CBDT keeps database of observed cases ,the action taken to
respond to those cases and results of the action.
 The policy engine must test and authenticate a waveform.
 Two main goals of policy engines
1) policy engine must be secure such that unauthorized
waveforms cannot be transmitted.
2) It must be liberal enough to allow many different types of
waveforms to run on the system.
 It is a component that translates between cognitive engine and
radio platform.
 When the cognitive engine wants to reconfigure the radio’s
waveform it uses generic communication theory representation
in XML.
 The mapping between the XML format to radio specific format
is done by parsing the XML file from cognitive engine and
formatting commands used to configure the radio.
Cognitive
engine
XML
implementa
tion file
Xml
C++
Parser python
java
Radio
platform
(SDR)
 The XML parser block is the translation block.
 The SDR control can also be accomplished by external
interface such as through HTTP, message passing etc.
 The radio framework used in this work is GNU radio software
radio.
 It has widely varying responsibilities depending on the
cognitive radio use case.
 Different instances
1) Control window
2) Simple configuration window
 In most idealist view of cognitive radio there is no user
interface.
 The important aspect of cognitive controller is its ability to use
many different implementations of the components described
above.
 It is configured through an XML file that defines which
components are currently attached.
 The cognitive controller can define and connect to multiple
sensors.
 Each component is described by a specific name that the
cognitive radio uses to identify when collecting the
information.
 Successful cognitive radios are aware ,can learn, and can take
action for any situation that might araise.These radios require
highly sophisticated learning and decision making capabilities.
 Techniques
1) Neural networks
2) HMM
3) Fuzzy logic
4) Evolutionary algorithms
5) Case based reasoning
 The cognitive engine concepts were introduced and its
implementation was shown. The major components of the
platform include sensors,optimizer,decision maker, policy
engine, radio framework, user interface. The discussion mostly
focused on defining the roles and responsibilities of each
component to provide the context from which to build a
cognitive radio. Various AI techniques were discussed.
 J.Mitola and G. Q. Maguire,Jr., “cognitive radio: making software
radios more personal,”IEEE proc. Personal communications,vol.
6,1999,pp.13-18.
 T.W.Rondeau, C.W.Bostian, D.Maldonado, A. Ferguson,
S.Ball,B.Le, and S.Midkiff,”cognitive radio in public safety and
spectrum management, "telecommunications policy and research
conference,vol.33,sep.2005
 FCC,”Implementing a Nation wide, Broadband , Interoperable public
safety network in the 700MHz band, "Federal communications
commision,Tech Rep. PS Docket No.06-229, Dec. 2006.
 http://scholar.lib.vt.edu/theses/available/etd-10052007081332/unrestricted/rondeau-dissertation.pdf

similar documents